Background Although the use of an intramedullary fibular allograft together with locking plate fixation can provide additional medial support and prevent varus malalignment in displaced proximal humeral fractures with promising results,the fibular autograft donor site often sustains significant trauma and cannot restore the articular surface of comminuted fractures.The aim of this study was to evaluate the clinical and radiographic outcomes of a locking plate and crest bone autologous graft for treating proximal humerus comminuted fractures.Methods We assessed the functional outcomes and complication rates in 40 patients with proximal humerus comminuted fractures.Eighteen patients were treated with a locking plate and an autologous crest bone graft (experimental group),and 22 were treated with only the locking plate and no bone graft (control group).Postoperative assessments included radiographic imaging,range of motion analysis,pain level based on the visual analogue scale (VAS),and the SF-36 (Short Form (36) Health Survey),as well as whether patients could retum to their previous occupation.Results All fractures healed both clinically and radiologically in the experimental group.There was no more than 2 mm collapse of the humeral head,and no osteonecrosis or screw penetration of the articular surface.In contrast,two patients had a nonunion in the control group,and they eventually accepted total shoulder replacements.The average time from surgery to radiographic union was significantly shorter in the experimental group ((4.66±1.63) months) compared with the control group ((5.98±1.57) months) (P ＜0.05).For the experimental versus controls groups,the mean shoulder active flexion (148.00±18.59 vs.121.73±17.20) degrees,extension (49.00±2.22 vs.42.06±2.06) degrees,intal rotation (45.00±5.61 vs.35.00±3.55)degrees,extal rotation (64.00±9.17 vs.52.14±5.73)degrees,and abduction (138.00±28.78 vs.105.95±15.66) degrees were all significantly higher (all P ＜0.001).The median SF-36 in the experimental group ((88.00±5.71) points) was significantly higher than that of the control group ((69.45±9.45) points; P ＜0.001).The median VAS pain level (mean rank,10.50) in the experimental group was lower than that (mean rank,47.19) of the control group (P ＜0.001).All but one patient (17 of 18,94.4％) in the experimental group reted to their previous activities or occupations,and that one patient changed to a different occupation because of slight restrictions to activities.On the other hand,four patients could not ret to their previous activities or occupations in the control group.Conclusion Locking plate fixation combined with an iliac crest bone graft is an effective technique for treating proximal humerus comminuted fractures.